Corrections

IMMUNOLOGY. For the article ‘‘Single-cell analysis of normal andFOXP3-mutant human T cells: FOXP3 expression withoutregulatory T cell development,’’ by Marc A. Gavin, Troy R.Torgerson, Evan Houston, Paul deRoos, William Y. Ho,Asbjørg Stray-Pedersen, Elizabeth L. Ocheltree, Philip D.Greenberg, Hans D. Ochs, and Alexander Y. Rudensky, whichappeared in issue 17, April 25, 2006, of Proc. Natl. Acad. Sci.

USA (103, 6659–6664; first published April 14, 2006; 10.1073�pnas.0509484103), the authors note the incorrect placement ofthe � in Fig. 1B. In addition, the position of Fig. 1E wasincorrect. The authors also note the omission of the � aftereach instance of ‘‘IFN-’’ in Fig. 3. The corrected figures andtheir legends appear below. These errors do not alter theconclusions of the article.

Fig. 3. Induced FOXP3 does not suppressIL-2 or IFN-� synthesis. (A) Freshly isolatedor stimulated (100 ng�ml anti-CD3) totalPBMC from donor 1745 were incubatedwith PMA, ionomycin, and monensin andstained for FOXP3 (rabbit IgG-digoxige-nin), IL-2, IFN-�, and surface markers as de-scribed in Materials and Methods. (B) Thepercentage of cytokine-expressing cellsamong FOXP3high, FOXP3low, or FOXP3�

cells is plotted. The distinction betweenhigh and low FOXP3 expression was notmade for CD4� cells and CD8� cells on day0. Data are representative of three sepa-rate experiments.

www.pnas.org�cgi�doi�10.1073�pnas.0603222103

Fig. 1. Flow cytometric detection of Foxp3 in murine and human cells. (A andB) Normal or Foxp3-deficient mouse lymph node cells were stained for Foxp3and cell-surface markers by using digoxigenin-conjugated mAb 3G3 (A) orFoxp3-specific rabbit antibody (B). CD4� gated lymphocytes are shown. (C–E)Normal (1792 and 1745) or FOXP3-deficient (IPEX) PBMC were stained forFOXP3 and lymphocyte markers by using digoxigenin-conjugated mAb 3G3(C) or digoxigenin-conjugated Foxp3-specific rabbit antibody (D and E). BothCD4� and CD8� gated lymphocytes are shown. Additional IPEX-1 PBMC werenot available for subsequent analysis with rabbit antibody. High backgroundstaining of Foxp3� cells is a consequence of the three-step staining procedure.

www.pnas.org PNAS � June 13, 2006 � vol. 103 � no. 24 � 9373–9374

CORR

ECTI

ON

S

Dow

nloa

ded

by g

uest

on

Aug

ust 2

0, 2

021

Dow

nloa

ded

by g

uest

on

Aug

ust 2

0, 2

021

Dow

nloa

ded

by g

uest

on

Aug

ust 2

0, 2

021

Dow

nloa

ded

by g

uest

on

Aug

ust 2

0, 2

021

Dow

nloa

ded

by g

uest

on

Aug

ust 2

0, 2

021

Dow

nloa

ded

by g

uest

on

Aug

ust 2

0, 2

021

Dow

nloa

ded

by g

uest

on

Aug

ust 2

0, 2

021

PHYSIOLOGY. For the article ‘‘Hypoxia-inducible myoglobin ex-pression in nonmuscle tissues,’’ by Jane Fraser, Luciane Vieirade Mello, Deborah Ward, Huw H. Rees, Daryl R. Williams,Yongchang Fang, Andrew Brass, Andrew Y. Gracey, and An-drew R. Cossins, which appeared in issue 8, February 21, 2006,of Proc. Natl. Acad. Sci. USA (103, 2977–2981; first publishedFebruary 9, 2006; 10.1073�pnas.0508270103), the authors notethat in the author line, the affiliations for Luciane Vieira deMello, Yongchang Fang, and Andrew Brass appeared incor-rectly, due to a printer’s error. In addition, the name YongchangFang should have appeared as Yongxiang Fang. The onlineversion has been corrected. The corrected author and affiliationlines and the original author footnotes appear below.

Jane Fraser*, Luciane Vieira de Mello*†, Deborah Ward*,Huw H. Rees*, Daryl R. Williams*, Yongxiang Fang‡,Andrew Brass‡, Andrew Y. Gracey*§,and Andrew R. Cossins*¶

*School of Biological Sciences, University of Liverpool, Liverpool L69 3BX,United Kingdom; ‡School of Biological Sciences, University of Manchester,Manchester M13 9PL,United Kingdom; and †Embrapa Recursos Geneticos eBiotecnologia, Brasılia, Brazil 70770-900

§Present address: Marine Environmental Biology, University of Southern California,Los Angeles, CA 90089-0371.

¶To whom correspondence should be addressed. E-mail: cossins@liv.ac.uk.

www.pnas.org�cgi�doi�10.1073�pnas.0602302103

GENETICS. For the article ‘‘Long-range multilocus haplotypephasing of the MHC,’’ by Zhen Guo, Leroy Hood, Mari Malkki,and Effie W. Petersdorf, which appeared in issue 18, May 2,2006, of Proc. Natl. Acad. Sci. USA (103, 6964–6969; firstpublished April 21, 2006; 10.1073�pnas.0602286103), the au-thors note that a grant was incorrectly cited. In line 5 of theAcknowledgments, ‘‘National Institutes of Health GrantsAI18029’’ should have read ‘‘National Institutes of HealthGrants CA18029.’’

www.pnas.org�cgi�doi�10.1073�pnas.0603506103

GENETICS. For the article ‘‘Insights into TOR function andrapamycin response: Chemical genomic profiling by using ahigh-density cell array method,’’ by Michael W. Xie, Fulai Jin,Heejun Hwang, Seungmin Hwang, Vikram Anand, Mara C.Duncan, and Jing Huang, which appeared in issue 20, May 17,2005, of Proc. Natl. Acad. Sci. USA (102, 7215–7220; firstpublished May 9, 2005; 10.1073�pnas.0500297102), the authorswish to add a reference to a paper by C. W. Xu (1), who employeda microarrayer to fabricate bacterial and yeast cell microarrayson nitrocellulose membrane.

1. Xu, C. W. (2002) Genome Res. 12, 482–486.

www.pnas.org�cgi�doi�10.1073�pnas.0603451103

9374 � www.pnas.org

Dow

nloa

ded

by g

uest

on

Aug

ust 2

0, 2

021

Hypoxia-inducible myoglobin expression innonmuscle tissuesJane Fraser*, Luciane Vieira de Mello*†, Deborah Ward*, Huw H. Rees*, Daryl R. Williams*, Yongxiang Fang‡,Andrew Brass‡, Andrew Y. Gracey*§, and Andrew R. Cossins*¶

*School of Biological Sciences, University of Liverpool, Liverpool L69 7ZB, United Kingdom; ‡School of Biological Sciences, University of Manchester,Manchester M13 9PL, United Kingdom; and †Embrapa Recursos Geneticos e Biotecnologia, Brasılia, Brazil 70770-900

Edited by George N. Somero, Stanford University, Pacific Grove, CA, and approved December 23, 2005 (received for review September 22, 2005)

Myoglobin (Myg) is an oxygen-binding hemoprotein that is widelythought to be expressed exclusively in oxidative skeletal andcardiac myocytes, where it plays a key role in coping with chronichypoxia. We now show in a hypoxia-tolerant fish model, that Mygis also expressed in a range of other tissues, including liver, gill, andbrain. Moreover, expression of Myg transcript was substantiallyenhanced during chronic hypoxia, the fold-change induction beingfar greater in liver than muscle. By using 2D gel electrophoresis, wehave confirmed that liver expresses a protein corresponding to theMyg-1 transcript and that it is significantly up-regulated duringhypoxia. We have also discovered a second, unique Myg isoform,distinct from neuroglobin, which is expressed exclusively in theneural tissue but whose transcript expression was unaffected byenvironmental hypoxia. Both observations of nonmuscle expres-sion and a brain-specific isoform are unprecedented, indicatingthat Myg may play a much wider role than previously understoodand that Myg might function in the protection of tissues from deephypoxia and ischemia as well as in reoxygenation and reperfusioninjury.

globin expression � hypoxia adaptation � microarray � proteomics

Myoglobin (Myg) has been intensively studied for manyyears in relation to protein structure (1). Myg is widely

thought to be exclusively expressed in oxidative myocytes ofskeletal and cardiac muscle, where it plays a role in oxygenstorage (2). Undoubtedly the richest source of Myg protein is inthe muscles of mammals such as whales and seals that undertakeextended breath-hold dives (3). However, Myg is also expressedin human oxidative muscle, and it has long been known that Mygexpression is up-regulated in response to increased demand foroxygen or during hypobaric hypoxia (4).

The wider functional role of Myg is still under discussion,because the transgenic knockout in mouse does not give rise tothe expected hypoxia-sensitive phenotype in adults because ofthe activation of alternative compensatory mechanisms (5). Inaddition to its well known role in oxygen storage and diffusion,Myg has also been implicated in protecting cell respiration fromNO-inhibition (6) and in scavenging reactive O2 species (ROS)(7). Given that ROS damage and NO synthesis occur morewidely than in muscle and that some tissues are as metabolicallyactive as muscle, some authorities have questioned why the tissuedistribution of Myg is not wider (7). This question was addressedby the more recent discovery of other members of the hemo-protein superfamily that may undertake these various roles,namely neuroglobin, cytoglobin (8), and globinX (9), consistentwith the evolution of ancient paralogues, which serve protectiveroles in tissues other than muscle. To date, however, members ofthe Myg subfamily have been restricted to muscle only.

The common carp, Cyprinus carpio, routinely experiencesextreme environmental hypoxia in isolated ponds, to which itdisplays a series of cardiorespiratory and metabolic adaptations(10). Together with other members of the Cyprinidae, C. carpiohas become a model for investigating fundamental mechanismsof hypoxia tolerance. We have undertaken a large-scale tran-

scriptomic screen of genes responding to hypoxia exposure in abroad range of carp tissues by using cDNA microarrays. Thisstudy has unexpectedly revealed the expression of Myg-1 inseveral nonmuscle tissues that we have now characterized.Second, we show that Myg expression is up-regulated in some ofthese tissues after hypoxia treatment. Finally, in screening cDNAlibraries from carp tissues, we have identified a previouslyuncharacterized Myg isoform called Myg-2, whose expression isrestricted to the neural tissue.

ResultsWe have explored responses to chronic hypoxia in the commoncarp, C. carpio, by using postgenomic screening techniques (11,12). EST characterization of a large collection of cDNAs frommultiple carp tissues (see carpBASE, http:��legr.liv.ac.uk) hasidentified cDNAs for a number of hemoproteins, including fourhemoglobin isoforms, one cytoglobin isoform, and a Myg iso-form that corresponded precisely to a known carp gene, heretermed Myg-1. The sequences of all of these genes aligned closelywith the corresponding known isoforms from fish homologues(Fig. 1a).

We have also discovered a second, Myg isoform, Myg-2(GenBank accession no. DQ338464). Despite a search for similarsequences in dbEST, GenBank, Swissprot, and the currentgenomic sequences for zebrafish, fugu, and Tetraodon, we failedto find a single hit. Myg-2 shares 78% sequence identity withMyg-1 and diverges particularly in two domains between residuenumbers 54–59 and 119–132 (Fig. 1b). The sequence differenceswere mapped onto the structure of tuna Myg (Fig. 2), showingthat none of the variable positions between the two carp isoformscontacted the bound heme.

By using cDNA microarrays containing 13,440 probes (13),including those for both Myg genes, we then explored transcriptexpression in carp exposed to hypoxia (0.8 mg O2�liter) over an8-day period at either 17°C or 30°C. These two temperaturesconstitute differing levels of hypoxic stress because of thediffering metabolic demands at the two temperatures. For liver,this analysis revealed a large number of responding genes, themost prominent of which was Myg (see Table 1, which ispublished as supporting information on the PNAS web site). Wehave also detected Myg transcript expression in brain, skeletalmuscle, and gill. In some tissues, Myg expression was hypoxia-

Conflict of interest statement: No conflicts declared.

This paper was submitted directly (Track II) to the PNAS office.

Abbreviation: Myg, myoglobin.

Data deposition: The sequence reported in this paper has been deposited in the GenBankdatabase (accession no. DQ338464). The microarray data files are available in the Array-Express database (accession no. E-MAXD-10).

See Commentary on page 2469.

§Present address: Marine Environmental Biology, University of Southern California,Los Angeles, CA 90089-0371.

¶To whom correspondence should be addressed. E-mail: cossins@liv.ac.uk.

© 2006 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0508270103 PNAS � February 21, 2006 � vol. 103 � no. 8 � 2977–2981

PHYS

IOLO

GY

SEE

COM

MEN

TARY

inducible; thus, in muscle, hypoxia was associated with a 4-foldincrease over 8 days of exposure but only at the higher, morestressful temperature (Fig. 3a). For liver, we found a muchstronger increase in expression at both temperatures (Fig. 3b).For gill, transcript amounts increased at both temperatures to a5-fold increase on day 5 (Fig. 3c). By contrast, brain wasunaffected at both temperatures (Fig. 3d), and cytoglobinshowed no changes in transcript expression in all tissues (data notshown).

Because of the possibility of cross-hybridization between thetwo types of cDNA probe, we tested the tissue-specificity of Mygisoform expression by designing PCR primers capable of distin-guishing between the two isoforms. We found that PCR productsfor Myg-1 were generated against cDNA from heart, kidney,liver, and gill and, to a much lesser extent, in brain (Fig. 4). Bycontrast, Myg-2 products were evident only in brain, whichcorresponds with the available ESTs for Myg-2 being isolatedfrom brain tissue. Therefore, we interpret the transcript expres-sion profiles as corresponding to Myg-1 for muscle, liver, and gilland both Myg-1 and Myg-2 for the brain.

Given that transcripts may not be translated, we soughtevidence for Myg protein expression (MYG) in nonmuscletissues by using proteomic techniques. Initially, we separatedprotein extracts of liver from hypoxia-treated carp on 2D PAGEand, with predictions of protein size and pI in conjunction withmass spectrometry, located the spot containing the MYG pro-tein corresponding to the Myg-1 isoform (Fig. 5a). This spot ondigestion and sequence analysis by MS�MS generated four longpeptide sequences that aligned perfectly against the predictedMYG-1 isoform (see Fig. 1b). We then undertook relativequantification of this spot by densitometry of silver-stained gelsof liver extracts from five carp specimens for both control andhypoxia-treated groups. The details in Fig. 5a indicate theconsensus differences in gel pattern, and Fig. 5b shows a 2.8-foldincrease in relative amounts of protein in the livers of 5-day,hypoxia-treated carp (Student’s t test, P � 0.05). We also usedan alternative quantification technique, namely iTRAQ (14),which directly compares the abundance of peptide fragmentsbetween control and 5-day-treated carp liver by using amine-specific peptide-tagging reagents. Again, this technique showeda 2.3-fold difference (Fig. 5c, P � 0.05).

DiscussionThe liver plays a key role in the hypoxia-adaptation of thecommon carp, primarily through the up-regulation of anaerobicmetabolism (15, 16). We now provide another potentially im-portant hypoxia-adaptive response in this tissue, namely thesubstantial up-regulation of MYG-1 protein expression in liverand other nonmuscle tissues through a transcriptional mecha-nism. Interestingly, gill showed induced responses at both 17°Cand 30°C, whereas brain and muscle showed induction only at thehigher temperature. This result might indicate increased sensi-tivity of gill tissue to hypoxic disturbance, despite the fact that thegill would be exposed to the highest environmental PO2s.

The expression of a liver Myg is unlikely to be unique to carpor even to members of the Cyprinidae family, because an ESTclone that aligns to Myg has been isolated from zebrafish kidneyand another from the liver of the minnow Fundulus (http:��genomics.rsmas.miami.edu�funnybase). Also, Myg transcript ex-pression has recently been detected in zebrafish gills as part ofa wider oligoarray screen (17), although the implications werenot explored. Even in the human, nonzero levels of Myg tran-

Fig. 1. Phylogenetic and sequence analysis of the two carp myoglobinisoforms. (a) Similarity analysis of the globin superfamily in fish. The source ofgene sequences included in this analysis and methods used are listed in Table2, which is published as supporting information on the PNAS web site. RTroutindicates rainbow trout. Bootstrapping confidence values (1,000 replicates)are shown for the major branches. In particular, the placing of both carpsequences reported here in the Myg group is supported at the highest boot-strapping level. (b) The amino acid alignment between carp MYG-1 andMYG-2 proteins with substituted residues indicated by a white backgroundand the peptide fragments identified in MYG-1 by using Q-TOF-MS by the solidlines and letters A–D.

Fig. 2. Diagram (made by using the program PYMOL), illustrating the surfaceformed by the residues substituted between carp MYG-1 and MYG-2 proteinssuperimposed on the crystal structure for tuna Myg (PDB code 1myt) (29).Selected tuna myoglobin residues are labeled as in the crystal structure. TunaMyg shares 73% sequence identity with cMyg-1 and 67% with cMyg-2. Thebound heme is shown with pink sticks in the center of the protein molecule.

2978 � www.pnas.org�cgi�doi�10.1073�pnas.0508270103 Fraser et al.

script have been detected on Affymetrix arrays in prostate, liver,and pancreatic tissue (Genecard, http:��genecards.bcgsc.bc.ca).Furthermore, Myg SAGE tags have been discovered in prostateand Myg ESTs in thymus, liver, prostate, and lung.

An important question is whether the nonmuscle expression ofthis gene is biologically meaningful; does such expression trulyrepresent expression within different cell types within the liver,gill, or brain, or might expression occur in smooth muscle or

endothelial cells within the tissue microvasculature (18), or doesthe expression merely represent slippage of transcriptional con-trol within tissue cells? In the liver of hypoxic carp, Mygconstitutes up to 1.5% of the protein extract applied to the 2Dgels and revealed by silver staining (see Fig. 5), which wecalculate to be equivalent to 1.2 mg of Myg per g of wet tissueweight. This value exceeds that of cardiac muscle of Antarcticfish species (0.4–0.7 mg�g) (19) but is somewhat lower than thatin the heart of the temperate Hemitripterus americanus (2.3mg�g) (20). Thus, although the very low levels of Myg transcriptin nonmuscle human tissues noted above may result fromslippage, carp liver expresses protein amounts that compare withthe ‘‘classical’’ tissue location for Myg, namely oxidative musclefibers. This finding, together with the clearly regulated andtissue-specific pattern of Myg expression during hypoxia, sup-ports the argument for a physiologically meaningful role.

Vascular smooth muscle cells do appear to express Myg (18).although the level of expression within smooth muscle cells andits precise cellular location remains uncharacterized. However,although we cannot exclude a smooth muscle component innonmuscle tissues, it seems unlikely to account for more than asmall proportion of what is an appreciable level of MYG proteinexpression, at least in liver. Moreover, we have observed majordifferences in transcript responses among the four tissue sampleswith respect to both fold-change and temperature. This tissue-specificity argues against a common mechanism, such as induc-tion in arteriolar smooth muscle, although we cannot exclude thetissues being different simply because of their different positionsin the circulatory loop and their different metabolic activities.Clarifying this point must await more precise definition of thecellular site(s) of protein expression in MYG-positive tissueswhen a suitable anti-carp MYG antibody becomes available.Also, understanding the physiological role of MYG expression,particularly in the brain and retina, must await analysis of theeffects of experimentally induced or ablated MYG expression ontolerance of hypoxia, reoxygenation, or reperfusion injury. In-terestingly, Nitta et al. (21) have recently shown that the virallyinduced expression of Myg in rat liver significantly reduced theextent of ischemia-reperfusion injury. This important resultdemonstrates that, under some circumstances, Myg can act inroles other than as an oxygen store and may be a crucial agentfor protection against the damaging effects of both oxygendeprivation and its subsequent restoration.

To our knowledge, the discovery of the carp Myg-2 gene is thefirst time that two Myg isoforms have been identified within asingle genome. Both genes are clearly members of the Mygsubfamily, and we show that the amino acid substitutions inMyg-2 relative to Myg-1 are not located in or close to the

Fig. 3. Changes in transcript expression for Myg-1 in muscle (a), liver (b), gill(c), and brain (d) during hypoxia exposure over an 8-day period at either 17°Cor 30°C. Values and bars represent the mean (�SEM), calculated relative to themean of the normoxia values measured on day 0 (n � 5). Statistical differencesof each time point relative to time 0 were calculated by using Student’s t test;

*, P � 0.05; **, P � 0.005; ***, P � 0.005. Please note the different ordinatescales for each of the four graphs.

Fig. 4. PCR products amplified from Myg-1 and Myg-2 transcripts in tissuesof hypoxia-treated carp by using isoform-specific primers (see Methods).Samples were analyzed by Tris-acetate-EDTA agarose gel electrophoresis toconfirm the presence of the expected 183-bp product.

Fraser et al. PNAS � February 21, 2006 � vol. 103 � no. 8 � 2979

PHYS

IOLO

GY

SEE

COM

MEN

TARY

heme-containing ligand-binding site. Instead, they are spreadover the surface of the protein, mainly in loops where they mighteither affect the conformational f lexibility of the critical hingeregion of the protein or, alternatively, form a different interfacewith an as yet unidentified protein or multiprotein complex.Distinguishing between these options will require detailed func-tional analysis.

This Myg2 isoform might be linked to the proposed whole-genome duplication event in carp within the last 12–15 millionyears (22) that has resulted in duplications in other key envi-ronmentally sensitive genes (23). If so, then this gene is likely tobe limited to a restricted clade within the Cyprinidae, includingCyprinus and perhaps members of the anoxia-tolerant genusCarassius (crucian carp and goldfish), which also possess dupli-cated genomes. The brain-specific expression of carp Myg-2points to promoter divergence in addition to divergence of thecoding sequence, and this divergence might have developed afterduplication through subfunctionalization of the ancestral pro-moter (24). In human brain and retina, neuroglobin appears toundertake roles attributed in muscle to Myg (25), and, in theretina, neuroglobin is expressed at the same high levels asobserved for Myg in muscle. This very high level of expressionis perhaps more consistent with a role in oxygen storage anddiffusion than in the more enzymatic roles of handling reactiveO2 species or NO. We have confirmed the existence of neuro-globin transcripts in carp brain by using PCR based on consensusfish neuroglobin motifs (data not shown). The existence ofMyg-2 in carp neural tissue seems, therefore, to augment ratherthan replace neuroglobin function.

In summary, we reveal the expression of the carp Myg-1 in somenonmuscle tissues, some of which display hypoxia-inducible expres-sion. We also show the existence of a second isoform, Myg-2,located exclusively in brain but which is not hypoxia-inducible.These unprecedented observations may account in part for thecarp’s hypoxia-tolerant physiology. Because nonmuscle Myg mayalso occur in the zebrafish, the phenomenon may well be morewidespread, and the widely assumed muscle-only expression of Mygshould not now be taken for granted. This interesting outcomejustifies the use of open-screening approaches in revealing unex-pected functional responses.

MethodsAnimals and Hypoxia Exposure. Common carp were acclimated toeither 17°C or 30°C for a 2-month period before hypoxia

exposure. On day 0, the oxygen level was lowered to 0.8 mg�literover a period of 3 hours, constituting a hypoxic stress of 8% and10% normoxia at 17°C and 30°C, respectively. We sampledtissues from five to eight fish (1000–1100 h) on days 0, 1, 3, 5, and8 after onset of hypoxia. Skeletal muscle was excised epaxiallyadjacent to the dorsal fin.

Microarray Experiment. For a full description of the array con-struction, cDNA sequencing, and annotation, refer to Gracey etal. (13). For the experiments reported here, we assessed a totalof 366 microarrays for two temperatures (17°C and 30°C), fivetime points, five tissues (brain, heart, muscle, liver, and gill), allsampled with 5-fold biological replication and with dye-swap.The data were subjected to quality control analysis as describedin ref. 13), normalized, and assessed statistically by using apopular signal-to-noise microarray statistic that uses permuta-tion of the sample labels to control for the false-discovery rateassociated with multiple testing procedures (26, 27). The rawmicroarray data files are available at ArrayExpress (accessionno. E-MAXD-10).

MYG-1 Protein Analysis. Liver was homogenized in buffer [40 mMTris base plus complete protease-inhibitor mixture (Roche Di-agnostics)] and the homogenate centrifuged (15,000 � g, 4°C, 30min). Supernatant proteins were precipitated by trichloroaceticacid-acetone, and resuspended in 320 �l of rehydration buffer [8M urea, 2 M thiourea, 4% 3-[(3-cholamidopropyl)dimethylam-monio]-1-propanesulfonate (CHAPS), 1% ASB14 detergent,0.2% biolytes, and 22 mM DTT]. Isoelectric focusing on pH 3–10linear strips was carried out at 250 V for 15 min, 10 kV for 3 h,and a linear ramp 10–60 kV�h. Strips were equilibrated (Bio-Rad Literature Bulletin no. 4006166, www.biorad.com) andproteins resolved on 12.5% polyacrylamide gels at 100 V for 1 hand 200 V for 5 h at 15°C. Gels were silver-stained (28) andscanned, and data were analyzed by using the program PDQUESTV7.1 (Bio-Rad). For relative quantification, the measured densityfor the MYG-1 spot has been expressed relative to the sum of alldetectable spots on the gel. Means were compared by usingStudent’s t test. Protein from four replicate fish from normoxicand 5-day-hypoxia-treated carp were each analyzed on triplicategels in two batches, each loaded with 150 �g of protein. Spots inthe predicted location for Myg were excised in-gel and digestedwith trypsin, and the extracted peptides were analyzed bynano-LC (Ultimate) MS�MS (Q-TOF micro, Waters). Proteins

Fig. 5. Proteomic identification of MYG-1 in carp liver. (a) The whole 2D gel and the two details to the right provide the consensus (n � 5) for the region ofthe Myg spot for both normoxic and hypoxia-treated carp. (b and c) The quantitative difference in expression of MYG-1 protein in extracts from normoxic and5-day-hypoxia-treated carp by using silver-stained densitometry of 2D gels (b) (n � 4) and by using the iTRAQ technique (c) (n � 3), respectively. For b, values(mean � SEM) represent fractional density of the MYG spot relative to the densitometric value for all spots detected on the entire gel, expressed as parts perthousand (ppt), whereas, in c, values (mean � SEM) represent intensities of mass spectrometric peaks derived from fragments obtained from normoxia- andhypoxia-treated specimens. For both b and c, the values for normoxia- and hypoxia-treatment samples were at a �0.05 level of significance (Student’s t test).The spot corresponding to an abundant liver protein, fatty acid-binding protein (FABP) is indicated for comparison.

2980 � www.pnas.org�cgi�doi�10.1073�pnas.0508270103 Fraser et al.

were identified by using the programs MASCOT and MASSLYNXPEPSEQ (Waters), searching against databases by using the FastSalgorithm. For relative quantification by iTRAQ, the 10- to30-kDa subfraction was first isolated by using spin concentrators,and the Myg in this subproteome was then analyzed by theiTRAQ MS�MS technique (http:��appliedbiosystems.com�pebiodocs�04350831). The quantitative values generated wereverified by using external protein standard mixtures.

RT-PCR. Isoform-specific PCR primers for Myg-1 and Myg-2 weredesigned to two variable regions between the Myg isoforms. For-ward, N-terminal primers were designed to the DNA sequencesencoding the N-terminal peptides MADHELV (Myg-1; atggccgat-cacgaactggtt) and MADYERF (Myg-2; atggctgattacgagcggttt). Re-verse primers were designed to the antisense DNA encoding thepeptide sequences between positions 54 and 61, NAAVKAHG

(Myg-1; gccgtgggccttcaccgctgcgtt) and DTLVASHG (Myg-2; ac-cgtgggacgccaccaacgtgtc). Oligo-dT-primed synthesis from 3 �g oftotal RNA from each tissue of hypoxia-exposed carp was per-formed by using Superscript II (Invitrogen) according to themanufacturer’s instructions. First-strand cDNA (1 �l; 1�20th of thetotal sample) served as template in a 50-�l PCR using 0.3 �l of Taqpolymerase (5 units��l; Bioline) and supplied 10� buffer, togetherwith the individual isoform-specific primers (0.5 �M each), dNTPs(0.2 mmol�liter), and MgCl2 (1.5 mmol�liter). Thermocyclingincluded one cycle at 95°C for 2 min, followed by 27 cycles of 95°Cfor 20 s, 60°C for 20 s, and 72°C for 1 min 30 s and, finally, oneextended polymerization step at 72°C for 4 min.

We thank Gregor Govan, Mark Prescott, and Michael Wilson for experttechnical help. This work was supported by grants from the NaturalEnvironment Research Council (U.K.) and the Biotechnology andBiological Sciences Research Council (U.K.).

1. Kendrew, J. (1963) Science 139, 1259–1266.2. Millikan, G. (1939) Physiol. Rev. 19, 503–523.3. Noren, S. & Williams, T. (2000) Comp. Biochem. Physiol. A 126, 181–191.4. Hoppeler, H. & Vogt, M. (2001) J. Exp. Biol. 204, 3133–3139.5. Godecke, A., Logel, U., Anger, K., Ing, Z., Irchenain, J., Ecking, U. &

Schrader, J. (1999) Proc. Natl. Acad. Sciences USA 96, 10495–10500.6. Wunderlich, C., Flogel, U., Godecke, A. & Schrader, J. (2003) Circ. Res. 92, 1352-.7. Brunori, M. (2001) Trends Biochem. Sci. 26, 209–210.8. Ascenzi, P., Bocedi, A., de Sanctis, D., Pesce, A., Bolognesi, M., Marden, M.,

Dewilde, S., Moens, L., Hankeln, T. & Burmester, T. (2004) Biochem. Mol.Biol. Education 32, 305–313.

9. Roesener, A., Fuchs, C., Hankeln, T. & Burmester, T. (2005) Mol. Biol. Evol.22, 12–20.

10. Stecyk, J. & Farrell, A. (2002) J. Exp. Biol. 205, 759–768.11. Gracey, A. Y., Troll, J. V. & Somero, G. N. (2001) Proc. Natl. Acad. Sci. USA

98, 1993–1998.12. Cossins, A. & Crawford, D. (2005) Nat. Rev. Genet. 6, 323–333.13. Gracey, A., Fraser, E., Li, W., Fang, Y., Taylor, R., Rogers, J., Brass, A. &

Cossins, A. (2004) Proc. Natl. Acad. Sci. USA 101, 16970–16975.14. Yan, W. & Chen, S. (2005) Brief. Funct. Genom. Proteom. 4, 27–38.15. Van Raaij, M. T. M., Van den Thillart, G. E. E. J. M., Vianen, G. J., Pit, D. S. S.,

Balm, P. H. M. & Steffens, A. B. (1996) J. Comp. Physiol. 166, 443–452.16. Zhou, B. S., Wu, R. S. S., Randall, D. J., Lam, P. K. S., Ip, Y. K. & Chew, S. F.

(2000) J. Fish Biol. 57, 1160–1171.

17. van der Meer, D., Van den Thillart, G., Witte, F., de Bakker, M., Besser, J.,Richardson, M., Spaink, H., Leito, J. & Bagowski, C. (2005) Am. J. Physiol.doi:10.1152.

18. Qiu, Y., Sutton, L. & Riggs, A. (1998) J. Biol. Chem. 273, 23426–23432.19. Moylan, T. & Sidell, B. (2000) J. Exp. Biol. 203, 1277–1286.20. Grove, T. & Sidell, B. (2002) Can. J. Zool. 80, 893–901.21. Nitta, T., Xundi, X., Hatano, E., Yamamoto, N., Uehara, T., Yoshida, M.,

Harada, N., Honda, K., Tanaka, A. & Sosnowski, C. (2003) J. Surg. Res. 110,322–331.

22. David, L., Blum, S., Feldman, M., Lavi, U. & Hillel, J. (2003) Mol. Biol. Evol.20, 1425–1434.

23. Polley, S. D., Tiku, P. E., Trueman, R. T., Caddick, M. X., Morozov, I. Y. &Cossins, A. R. (2002) Am. J. Physiol. 284, R41–R50.

24. Force, A., Lynch, M., Pickett, F. B., Amores, A., Yan, Y. L. & Postlethwait, J.(1999) Genetics 151, 1531–1545.

25. Burmester, T., Weich, B., Reinhardt, S. & Hankeln, T. (2000) Nature 407,520–523.

26. Tusher, V. G., Tibshirani, R. & Chu, G. (2001) Proc. Natl. Acad. Sci. USA 98,5116–5121.

27. Benjamini, Y. & Hochberg, Y. (1995) J. R. Stat. Soc. B 57, 289–300.28. Yan, J., Wait, R., Berkelman, T., Harry, R., Westbrook, J., Wheeler, C. &

Dunn, M. (2000) Electrophoresis 21, 3666–3672.29. DeLano, W. L. (2003) PYMOL, Ver. 0.97 (DeLano Scientific, San Carlos, CA).

Fraser et al. PNAS � February 21, 2006 � vol. 103 � no. 8 � 2981

PHYS

IOLO

GY

SEE

COM

MEN

TARY